Download One-Stage Amplifier Design Consideration of a

AN-1439
APPLICATION NOTE
One Technology Way • P.O. Box 9106 • Norwood, MA 02062-9106, U.S.A. • Tel: 781.329.4700 • Fax: 781.461.3113 • www.analog.com
One-Stage Amplifier Design Consideration of a High Fidelity System
by Roy He
INTRODUCTION
Therefore, one-stage design method is more suitable for
scenarios that require high performance to price ratio products,
such as mobile phones.
With an increasing demand of built in, high fidelity systems for
portable devices, for example, mobile phones and active
headphones, amplifiers for larger driving power are widely
implemented into high fidelity systems.
Figure 2 shows the typical connection diagram of a high fidelity
system.
A high fidelity design is a system level design where each device has
its own functions and performs together as a whole. The typical
desired specification of codec or digital-to-analog converter (DAC)
in portable devices is usually indicated as follows:
As part of a high fidelity system, the output amplifier is a
critical element for both objective and subjective performances
of the whole audio performance by providing ample output
power toward its load, the headphones. Analog Devices, Inc.,
offers a wide range of low noise, high performance amplifiers
for such applications, like the ADA4841-2, ADA4896-2,
ADA4075-2, AD8599, AD8610, AD8397, and ADA4807-2.
These amplifiers are widely used in a high fidelity related
system circuit design.
•
•
Voltage swing up to 2 V rms.
ROUT is low enough to achieve optimal performance and to
be ignored, which helps attain better damping coefficient
and output amplitude levels.
Differential voltage output.
•
Because there are various methods for high fidelity system
design, each scenario cannot be described in this application
note. This application note discusses the one-stage amplifier
structure, which is low cost, easy to design, and widely used by
mobile phone companies.
The simple and low cost typical circuit is shown in Figure 1.
In Figure 1, VAC is the audio signal and VDC is the internal bias
from the audio DAC.
C1
ONE-STAGE SIMPLE AMPLIFIER HIGH FIDELITY
DESIGN
R1
R2
ROUT
VAC
VDC
The amplifier block in Figure 2 shows there are different kinds
of circuit designs based on the general audio amplifier that
amplify, filter, buffer, and signal condition. When compared
with a two-stage or three-stage amplifier design, a one-stage
design method solution provides less components in the design,
such as lower cost and a simplified printed circuit board (PCB)
layout, while still providing excellent audio performance.
RS
VAC
ROUT
HEADPHONE
OUTPUT
R3
R4
AUDIO DAC
C2
15221-002
For better understanding of the high fidelity system structure as
a whole, the amplifier design is discussed in this section.
Figure 1. Diagram of Typical One-Stage Simple Amplifier Structure
HEADPHONES
ANALOG LEFT CHANNEL
DIGITAL AUDIO INTERFACE
CODEC
OR
DAC
AMPLIFIER
ANALOG RIGHT CHANNEL
Figure 2. Typical Connection Diagram of a High Fidelity System
Rev. 0 | Page 1 of 6
15221-001
FRONT-END
DSP
AN-1439
Application Note
TABLE OF CONTENTS
Introduction ...................................................................................... 1
Output Series Resistor ..................................................................5
One-Stage Simple Amplifier High Fidelity Design ...................... 1
Bandwidth Setting .........................................................................5
Revision History ............................................................................... 2
Noise Performance ........................................................................5
Calculation Guidance and Design Parameters ............................. 3
Gain Configuration .......................................................................5
Calculation .................................................................................... 3
Crosstalk .........................................................................................5
Design Consideration .................................................................. 4
Dynamic Range (DNR) ................................................................5
Debugging Methods and Analysis ................................................. 5
Filter Design .......................................................................................6
Shielding ........................................................................................ 5
REVISION HISTORY
12/2016—Revision 0: Initial Version
Rev. 0 | Page 2 of 6
Application Note
AN-1439
CALCULATION GUIDANCE AND DESIGN PARAMETERS
CALCULATION
Then,
The standard testing method of a high fidelity audio system
performance includes subjective and objective tests.
RTINOISE(f) ≈
In a subjective hearing test, the performance is due to the standard
subjective hearing style of a human in regards to how they feel
while music is played, which is not easy to calculate.
The audio DAC setting and the circuit design component values
can be tuned based on the various expected hearing styles. The
focus of this application note is only on how to improve objective
performance by tuning circuit design at the hardware level.
Among plenty of test items, total harmonic distortion plus noise
(THD plus N) are important factors for objective performance
tests.
4 KTRP1  4 KTRP 2  E N 2  I N 2  RP12  RP 2 2 
where:
RTINOISE is the referred to input noise, in V/√Hz.
K is the Boltzmann’s constant, = 1.3806505(24) × 10−23 J/K.
T is the temperature.
RP1 and RP2 are the parallel resistor values.
EN is the voltage noise spectral density.
IN is the current noise spectral density.
 R2 
RTONOISE(f) = RTI NOISE  1  
 R1 
Typically, for small signal levels, N is the dominant factor,
which is deduced in the following calculations based on the
indicated circuit in Figure 2. The capacitor has little effect on
the formula, which is why the capacitor effect is ignored.
where:
RTONOISE is the referred to output noise, in V/√Hz.
R2 is the feedback resistor, shown in Figure 1.
R1 is the series input resistor, shown in Figure 1.
Assume the following statements:
RTORMS =


EN and IN are the voltage and current noise spectral density
of the one-stage amplifier shown in Figure 1.
Frequency cutoff (fC) is the system bandwidth.
To make the circuit more symmetrical, select the resistor values as
R1 = R3, R2 = R4
(1)
1 fC
 0 RTONOISE ( f ) ≈ RTONOISE × 1.57  f C
2
(2)
(3)
where:
RTORMS is the rms value of the referred to output noise.
f is the frequency.
RP1 = R1 || R2
THD + N = 20log
RP2 = R3 || R4
RTONOISE  THD
R2
VIN   
 R1 
(4)
where VIN is the input voltage.
See Equation 5 to calculate THD + N.
THD + N (dB) ≈ 20log
R2 
4 KTRP1  4 KTRP2  EN 2  I N 2  R2P1  R2P 2    1 
  1.57  f C  THD
R1 

R2
VIN   
 R1 
Rev. 0 | Page 3 of 6
(5)
AN-1439
Application Note
DESIGN CONSIDERATION
For a small signal level scenario, THD is small enough to be
ignored because it is almost equal to zero, which has no effect
on the system due to the electrical characteristics of an integrated
circuit (IC) design. Thus, noise is the dominant factor affecting
the THD + N results under such small signal level situations.
Taking the ADA4807-2 as an example, the approximate noise
performance within a 22 kHz bandwidth is calculated if R1 = R3 =
1 kΩ and R2 = R4 = 1 kΩ. Figure 3 shows the calculated result
using the mathematic tool, wxMaxima.
The noise result is improved when using test equipment like the
SYS-2712 or the APx555 to analyze audio in dBA because the
equipment adds an A weighted filter to reduce noise, which
simulates the response to sound of the human ear.
Usually, THD is much more serious if the desired load power
exceeds the limit for output of any amplifier, which means,
unlike in light load conditions, THD must be carefully
considered, especially in heavy load scenarios.
Therefore, voltage and current noise, bandwidth, resistors, output
capability, and gain setting affects objective test results at the
hardware level.
As a result, users can choose and judge related parameters, as
shown in Figure 1 and Figure 3.
Moreover, other parameters, like slew rates, common-mode
rejection ratio (CMRR), power supply rejection ratio (PSRR), and
voltage swing, are system factors that must be taken into
consideration during audio system design.
15221-003
To improve performance, power, noise, and electromagnetic
interference (EMI) must be taken into design consideration as
well. Audio performance is sensitive and easily affected by WiFi or
radio frequency (RF) environments; therefore, good shielding and
layout are required.
Figure 3. Calculating Results of the ADA4807-2 with wxMaxima
Rev. 0 | Page 4 of 6
Application Note
AN-1439
DEBUGGING METHODS AND ANALYSIS
SHIELDING
GAIN CONFIGURATION
A shielding case to cover a high fidelity audio circuit is highly
recommended. Layout must be carefully treated to avoid possible
interference from other parts of the whole circuit system. The
layout is critical because the audio performance is related to
signal integrity and electrical environment, for instance, strong
interference appears around dc to dc parts or a RF power amplifier.
Proper layout, including trace routing and PCB stackup, is expected
to help avoid noise and distortion.
To achieve lower noise generated by the amplifier, calculate the
proper gain setting with a resistor value. Taking the circuit shown
in Figure 1 as an example, assume the load is RL, VL is the voltage
load, and IL (loads not shown in Figure 1) is the current load,
therefore
OUTPUT SERIES RESISTOR
where:
VDAC is the voltage output from the DAC.
RS is the output series resistance, shown in Figure 1.
RL is the load resistance.
An output series resistor, shown as RS in Figure 2, is
recommended for the following reasons:
•
Guarantees the stability of the amplifier and covers
varieties of headphone loads.
Prevents the output power from exceeding the limit of the
amplifier.
Tunes for output signal amplitude, as well as subjective
hearing.
•
•
VL =
IL =
VDAC × Gain × RL
(6)
RS + R L
VDAC × Gain
(7)
RS + RL
For given RL and VDAC values, and the required VL and IL values,
calculate and set the optimal gain, as well as the desired fC, with
the proper capacitor value according to the design requirement.
BANDWIDTH SETTING
CROSSTALK
Theoretically, a narrower bandwidth results in lower noise while
limiting the flatness of frequency response, thus requiring a tradeoff point. Typically, the best cutoff area range is from 60 kHz to
200 kHz.
Crosstalk indicates the isolation between audio channels.
Impedance between different ground nets of each channel
contributes to most of the crosstalk. The layout principle can
achieve low impedance connection between ground networks. One
thing worth noting is that different kinds of ear jacks result in
different GND connection impedances, which matters in crosstalk
performance because ear jacks with a lower impedance connection
helps to achieve better performance.
NOISE PERFORMANCE
0
–10
–20
–30
–40
–50
–60
–70
–80
–90
–100
–110
–120
–130
–140
–150
–160
DYNAMIC RANGE (DNR)
As discussed in the Calculation Guidance and Design Parameters
section, for small signal, noise (N) is the dominant factor that
affects THD + N results. DNR is tested with almost no THD, only
the noise condition. There is almost no THD in Figure 5; therefore,
N is the main factor that affects DNR results. A lower noise floor
results in improved DNR.
–50
–60
–70
100
1k
10k
FREQUENCY (Hz)
30k
Figure 4. Floor Noise Amplitude Level vs. Frequency
–80
–90
–100
–110
LEFT AUDIO CHANNEL
RIGHT AUDIO CHANNEL
–120
–130
–140
–150
–160
–170
20
100
1k
FREQUENCY (Hz)
Figure 5. Dynamic Range vs. Frequency
Rev. 0 | Page 5 of 6
5k
15221-005
20
DYNAMIC RANGE (dBrA)
LEFT AUDIO CHANNEL
RIGHT AUDIO CHANNEL
15221-004
FLOOR NOISE AMPLITUDE LEVEL (dBV)
From an analog design perspective, a larger gain results in larger
noise; therefore, for a one-stage amplifier, minimal gain must be set
to meet the output requirements. Either poor power or EMI design
can result in noise spurs at certain areas, as shown in Figure 4.
AN-1439
Application Note
FILTER DESIGN
–4
–100
–32
–120
–36
–140
–40
–160
–44
–180
Figure 7. Amplitude and Phase vs. Frequency
60
50
PHASE (Degrees)
+5V
40
30
20
OUT
10
R1
1.6kΩ
R4
1.6kΩ
–200
1M
100k
–5V
C1
750pF
0
0
Figure 6. Schematic Diagram of a Bessel Active Filter
Table 1 indicates the different phase drifts and linearity at various
frequencies. The linearity curve is shown in Figure 8.
A one-stage amplifier high fidelity model requires system level
design consideration, which includes headphone load, a DACcompatible interface, audio power design, layout considerations,
and system level tuning.
10k
15k
20k 25k 30k 35k
FREQUENCY (Hz)
40k
45k
50k
Figure 8. Phase vs. Frequency
Figure 7 shows that the phase drift is in excellent linearity within
the 20 Hz to 20 kHz, which is typically known as the human
audible frequency range. The simulation of amplitude and phase
vs. frequency results are shown in Figure 7.
If the filter design, as shown in Figure 1, is changed to a secondorder filter, the calculation of noise must change.
5k
For a more specific scenario design, compatibility of headphone
parameters, such as frequency, impedance, sensitivity, cumulative
spectral decay (CSD), and features, are also important for the entire
audio system design. Audio designers can tune the analog
components, like the amplifier, resistor, and capacitor, as well as
the digital parameter setting like the digital filter and oversampling
configuration, to achieve the final optimized performance.
To calculate the linearity listed in Table 1 and the slope shown
in Figure 8, use the following equation:
Linearity = Phase × 1000 ÷ Frequency
Table 1. Simulation Result of Filter Linearity
Parameter
|Phase|
Linearity
100
0.130
1.299
1000
1.298
1.298
5000
6.492
1.298
15221-008
C3
2.25nF
100
1k
10k
FREQUENCY (Hz)
70
15221-006
DAC_OUT+
10
PHASE (Degrees)
–80
–28
15221-007
AMPLITUDE (dB)
–60
–24
1
C4
750pF
U1
R25
1.6kΩ
–40
–20
–48
R6
1.6kΩ
DAC_OUT–
–20
PHASE RESPONSE
–16
Figure 6 illustrates one example of a Bessel active filter schematic
diagram. DAC_OUT+ and DAC_OUT− are the output signals
from the DAC, which are also the positive and negative inputs
of the Bessel filter.
R3
1.6kΩ
0
–12
Amplifiers can easily result in phase drift, which means different
signal frequencies lead to different time delays after passing
through the amplifier, which is also known as group delay. Bessel
filters can guarantee linear phase performance and achieve the
same group delay towards different signal frequencies.
R2
1.6kΩ
20
AMPLITUDE
–8
Alternatively, the filter design can be refined with a second-order
active filter design, such as a Butterworth, Chebyshev, Elliptic, or
Bessel filter. Different types of filters hold different characteristics;
therefore, a user must choose the specific filter type according to
the specific requirement of the whole system. This section discusses
the Bessel linear phase filter.
C2
2.25nF
40
0
Figure 1 shows a typical one-stage audio amplifier circuit design,
using a simple first-order active filter, thereby achieving a limited
filter effect.
Frequency (Hz)
10,000
20,000
12.984
25.947
1.298
1.297
©2016 Analog Devices, Inc. All rights reserved. Trademarks and
registered trademarks are the property of their respective owners.
AN15221-0-12/16(0)
Rev. 0 | Page 6 of 6
30,000
38.794
1.293
50,000
63.245
1.265
Unit
Degrees
m°/Hz